Process modeling and optimization of sorrel biodiesel synthesis using barium hydroxide as a base heterogeneous catalyst: appraisal of response surface methodology, neural network and neuro-fuzzy system

In this study, three different modeling tools, viz. response surface methodology (RSM), artificial neural network (ANN) and adaptive neuro-fuzzy inference system (ANFIS), were used to model the process of conversion of sorrel (Hibiscus sabdariffa) oil to H. sabdariffa methyl esters (HSME). The high free fatty acid (13.47%) of the sorrel oil was reduced to 0.62 ± 0.05% using methanol/oil molar ratio of 40:1, catalyst (ferric sulfate) weight of 15 wt%, reaction time of 3 h and temperature of 65 °C, followed by transesterification step. The developed models for the transesterification process were all found to be reliable and accurate when subjected to different statistical tests. ANFIS model [coefficient of determination (R²) = 0.9944] was better than ANN model (R² = 0.9875), while RSM model (R² = 0.9789) was the least accurate. The results of process optimization for the transesterification showed that genetic algorithm (GA) performed better than RSM. The highest HSME yield of 99.71 wt% could be obtained under optimal condition of methanol/oil molar ratio 8:1, catalyst weight 1.23 wt% and reaction time 43 min while keeping temperature at 65 °C using ANFIS model which has been optimized with GA. The sensitivity analyses showed that time was the most important input variable, followed by methanol/oil molar ratio and lastly catalyst weight. Quality characterization of the HSME showed that it could serve as an alternative to petro-diesel.